Simulation of restricted probesets for spatial transcriptomics
For probe-based spatial transcriptomics, the number and identity of genes that have probesets is limited. This script queries how this restricted gene set might perform in cell identification by routine clustering.
Environment
library(future)
plan("multiprocess", workers = 4)
Error: No such strategy for futures: ‘multiprocess’Read in IPF Cell Atlas (GSE136831)
This notebook depends on the output of the read-in script. for convenience, can bypass script by loading stored Seurat object directly.
# This is more reliable
#source(knitr::purl("IPF-Cell-Atlas__read-in.Rmd", quiet=TRUE))
knitr::knit("IPF-Cell-Atlas__read-in.Rmd", output = tempfile())# faster
load('ipfatlas.cells_fromd8d92b9.Robj')Simulate Xenium lung panel 1
This is the panel used in the initial trial runs with the genomics CoLabs.
xenium.panel1.genes <- read.csv('Xenium_panel_order_08_10_23.csv', header = TRUE)$Geneipfatlas.xen1panel.cells <- subset(ipfatlas.cells, features=xenium.panel1.genes)Some genes in the xenium panel are not in the IPF dataset. Not sure why the dataset in GSE136831 is missing these genes. so actual list is a bit shorter. These are the missing genes:
setdiff(xenium.panel1.genes, rownames(ipfatlas.xen1panel.cells))Standard clustering
print('hello')[1] "hello"ipfatlas.xen1panel.cells <- NormalizeData(ipfatlas.xen1panel.cells)Normalizing layer: counts
Performing log-normalization
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.xen1panel.cells <- FindVariableFeatures(ipfatlas.xen1panel.cells)Finding variable features for layer counts
Calculating gene variances
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.xen1panel.cells <- ScaleData(ipfatlas.xen1panel.cells, features=rownames(ipfatlas.xen1panel.cells))Centering and scaling data matrix
|
| | 0%
|
|======================================================================================================================| 100%ipfatlas.xen1panel.cells <- RunPCA(ipfatlas.xen1panel.cells)PC_ 1
Positive: ELF3, AGR3, ENAH, PDE4D, TACSTD2, CLDN4, CDH1, AGR2, TMC5, EPCAM
RORA, EGFR, EHF, KRT8, CD24, TC2N, CTTN, CD247, KLF5, NTN4
KRT7, MALL, FOXJ1, TMPRSS2, CCDC78, MTUS1, F3, GZMB, SFTA2, TFF3
Negative: AIF1, MARCO, VSIG4, S100A9, MS4A4A, CTSL, MCEMP1, GLIPR2, SLC1A3, FCGR3A
CSTA, S100A8, LGALS3, RETN, ANPEP, CD86, KLF4, TREM2, CD163, SYK
ADAM17, IFI6, NCEH1, CD14, FCGR1A, HAVCR2, AQP9, CLEC12A, LILRB4, MIS18BP1
PC_ 2
Positive: CD247, RUNX3, STAT4, GZMB, NKG7, GZMA, CD3D, CD2, KLRB1, CD3E
KLRD1, LCK, CD8A, IQGAP2, GPR171, SAMD3, KLRC1, IFNG, CD8B, PIM2
TBX21, CTLA4, GATA3, CD28, FGFBP2, CXCR6, CD40LG, TNFAIP3, LAG3, TNFRSF18
Negative: ELF3, AGR3, CDH1, CLDN4, TACSTD2, TMC5, AGR2, KRT8, LGALS3, EPCAM
EHF, ENAH, MTUS1, KLF5, KRT7, LGALS3BP, EGFR, CTTN, ATP1B1, CD24
MYO6, MALL, NCEH1, SFTA2, TMPRSS2, SEMA3C, F3, EGLN3, CCDC78, NTN4
PC_ 3
Positive: AGR3, ELF3, CDH1, TMC5, AGR2, CLDN4, TACSTD2, EPCAM, EHF, KRT8
CCDC78, FOXJ1, CD24, TMPRSS2, SFTA2, KRT7, KLF5, SFTPD, ALOX15, GKN2
CEACAM6, MAP7, DUOX1, IQGAP2, MUC16, FAM184A, BCAS1, ITGB6, CP, ATP1B1
Negative: SFRP2, COL1A1, COL5A2, THY1, FBN1, GNG11, CRISPLD2, THBS2, MEDAG, CD34
POSTN, SLIT3, SCN7A, PLA2G2A, RAMP2, CTHRC1, ADAMTS1, INMT, PDGFRB, NID1
PDPN, APOD, CLDN5, IL33, DCLK1, LTBP2, MFAP5, SVEP1, IGF1, TMEM100
PC_ 4
Positive: CD247, CD3D, LGALS3BP, GZMB, GZMA, RBP4, CD2, CD3E, PCOLCE2, KLRD1
IL7R, NKG7, IQGAP2, KLRB1, LCK, FCGR3A, CD8A, ATP1B1, RUNX3, NCEH1
FGFBP2, SAMD3, MCEMP1, VSIG4, FABP3, STAT4, GPI, CD8B, MARCO, IFI6
Negative: FCN1, S100A12, ADAM28, GPR183, CD300E, HIF1A, VEGFA, TNFRSF13C, AREG, MS4A1
CD79A, CSF3R, BANK1, CLDN5, CD14, S100A8, SERPINB2, LILRB2, LILRA5, CLEC10A
RAMP2, MET, SELL, SMAD3, VWF, CLEC4E, TNFRSF13B, SHANK3, CD1C, FCER1A
PC_ 5
Positive: CLDN5, RAMP2, VWF, SHANK3, GNG11, ADGRL4, ACKR1, MMRN1, FCN3, PLVAP
TMEM100, IL33, TM4SF18, KDR, SMAD6, SELP, ITGA6, TFF3, POSTN, LYVE1
SELE, ICA1, MTUS1, PROX1, CA4, RND1, APOLD1, MYO6, STC1, CD34
Negative: COL1A1, SFRP2, CRISPLD2, COL5A2, THY1, THBS2, SCN7A, PLA2G2A, SLIT3, DCLK1
PDGFRB, SVEP1, MEDAG, MFAP5, LTBP2, GPR183, CXCL14, RARRES1, FCN1, AREG
VEGFA, INMT, TSPAN8, COL8A1, PI16, APOD, PAMR1, FBN1, S100A12, CD300E ipfatlas.xen1panel.cells <- RunUMAP(ipfatlas.xen1panel.cells, dims=1:30)Warning: The default method for RunUMAP has changed from calling Python UMAP via reticulate to the R-native UWOT using the cosine metric
To use Python UMAP via reticulate, set umap.method to 'umap-learn' and metric to 'correlation'
This message will be shown once per session22:57:20 UMAP embedding parameters a = 0.9922 b = 1.112
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
22:57:20 Read 253424 rows and found 30 numeric columns
22:57:20 Using Annoy for neighbor search, n_neighbors = 30
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
22:57:20 Building Annoy index with metric = cosine, n_trees = 50
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
22:58:04 Writing NN index file to temp file /tmp/RtmpvuYL8i/filef3dd462abf0f
22:58:04 Searching Annoy index using 1 thread, search_k = 3000
23:00:24 Annoy recall = 100%
23:00:25 Commencing smooth kNN distance calibration using 1 thread with target n_neighbors = 30
23:00:41 Initializing from normalized Laplacian + noise (using RSpectra)
23:01:17 Commencing optimization for 200 epochs, with 12124680 positive edges
Using method 'umap'
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
23:03:40 Optimization finishedDimPlot(ipfatlas.xen1panel.cells, group.by='CellType_Category')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.xen1panel.cells, group.by='Manuscript_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.xen1panel.cells, group.by='Subclass_Cell_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.xen1panel.cells, group.by='Subject_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.xen1panel.cells, group.by='Disease_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
Subset epithelium
ipfatlas.xen1panel.epi.cells <- subset(ipfatlas.xen1panel.cells, subset=CellType_Category=='Epithelial')ipfatlas.xen1panel.epi.cells <- NormalizeData(ipfatlas.xen1panel.epi.cells)Normalizing layer: counts
Performing log-normalization
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.xen1panel.epi.cells <- FindVariableFeatures(ipfatlas.xen1panel.epi.cells)Finding variable features for layer counts
Calculating gene variances
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.xen1panel.epi.cells <- ScaleData(ipfatlas.xen1panel.epi.cells, features=rownames(ipfatlas.xen1panel.epi.cells))Centering and scaling data matrix
|
| | 0%
|
|=====================================================================================================================| 100%ipfatlas.xen1panel.epi.cells <- RunPCA(ipfatlas.xen1panel.epi.cells)Warning: The following 1 features requested have zero variance; running reduction without them: ICA1PC_ 1
Positive: SFTA2, SFTPD, PEBP4, ABCA3, GKN2, CSF3R, MALL, F3, ITGB6, ALOX15B
FASN, ETV5, AREG, DAPK2, COL8A1, PLA2G4F, ERN1, HIF1A, RND1, MYC
DMBT1, STEAP4, CFTR, AGER, VEGFA, NFKB1, S100A9, TMPRSS2, HP, MET
Negative: CCDC78, FOXJ1, CD24, SEMA3C, ENAH, ALOX15, POU2AF1, TFF3, BCAS1, MUC16
FGFBP2, TMC5, TP73, FOXN3, CP, IQGAP2, SLC2A1, CCNA1, RARRES1, PAMR1
RGS5, CD4, ITGB4, AGR3, IFI6, SVEP1, CYP2F1, COL5A2, APOD, EPCAM
PC_ 2
Positive: PDPN, PLA2G2A, AGER, UPK3B, MEDAG, SCEL, WT1, VCAM1, COL8A1, SEMA3B
THBS2, INMT, SLIT3, GNG11, CRISPLD2, DAPK2, RBP4, DES, PTGS1, WFS1
CD34, VEGFA, AQP9, KRT5, TP63, MMP1, PCOLCE2, IL1RL1, POSTN, DIRAS3
Negative: AGR2, ELF3, CLDN4, AGR3, XBP1, LGALS3, TACSTD2, BAIAP2L1, EPCAM, KRT8
TMC5, S100A9, EHF, STEAP4, TSPAN8, CP, CDH1, TFF3, SEC11C, MTUS1
PRDX6, MALL, MUC5B, ATP1B1, KLF5, MET, LTF, FOXN3, GDF15, CYP2F1
PC_ 3
Positive: PDE4D, ABCA3, SFTPD, AGR3, CSF3R, PEBP4, GKN2, CCDC78, SFTA2, ALOX15B
FAM184A, TMC5, PLA2G4F, FOXJ1, CD38, FASN, ELF3, TMPRSS2, NTN4, DUOX1
ETV5, RND1, GLCCI1, TP73, POU2AF1, EGLN3, DMBT1, EPCAM, CDH1, TC2N
Negative: PLA2G2A, KRT17, KLK11, MEDAG, CXCL6, KRT5, PDPN, VCAM1, KRT15, WT1
TSPAN8, IGFBP3, ADAM28, UPK3B, GNG11, PIM1, COL1A1, RARRES1, CRISPLD2, KLF4
SLIT3, RORA, SERPINB2, RBP4, HP, KRT7, THBS2, KDR, LTF, MUC5B
PC_ 4
Positive: KRT17, KRT15, KRT5, KRT7, ADAM28, CEACAM6, TSPAN8, TP63, MMP1, KLK11
CDH1, IL33, MUC5B, TACSTD2, DAPK2, SLC15A2, TRPC6, F3, ERN2, CXCL6
CXCL14, LTF, ANPEP, DCLK1, EHF, ITGB4, ELF3, CSTA, AGER, SPDEF
Negative: PLA2G2A, MEDAG, VCAM1, WT1, HP, GNG11, SLIT3, THBS2, RBP4, INMT
CRISPLD2, ALOX15, CD34, SEMA3C, PDPN, COL8A1, SVEP1, ADAMTS1, CCDC78, AQP9
RARRES1, MUC16, UPK3B, FOXJ1, DES, PRDX6, EGLN3, TNFAIP3, COL5A2, PCOLCE2
PC_ 5
Positive: KRT5, TP63, KRT17, KRT15, ITGA6, MMP1, PDE4D, MET, CSF3R, MTUS1
ETV5, EHF, HIF1A, TRPC6, NFKB1, IL33, ERN1, RORA, PLA2G4F, ADAM17
ALOX15B, AREG, MYC, ENAH, POSTN, IL4R, ADAMTS1, TNFAIP3, CXCL14, TGFB1
Negative: AGER, CEACAM6, SCEL, WFS1, KRT7, SEMA3B, VEGFA, UPK3B, DAPK2, COL8A1
SFTA2, LGALS3BP, IFI6, EPCAM, TACSTD2, KLK11, KRT8, AGR3, SLC15A2, LGALS3
PDPN, PARP14, CD24, PRDX6, OTUD7B, DCLK1, TSPAN8, MUC5B, GLCCI1, TFF3 ipfatlas.xen1panel.epi.cells <- RunUMAP(ipfatlas.xen1panel.epi.cells, dims=1:30)23:49:18 UMAP embedding parameters a = 0.9922 b = 1.112
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
23:49:18 Read 13370 rows and found 30 numeric columns
23:49:18 Using Annoy for neighbor search, n_neighbors = 30
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
23:49:18 Building Annoy index with metric = cosine, n_trees = 50
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
23:49:19 Writing NN index file to temp file /tmp/RtmpvuYL8i/filef3dd2d8fb7a6
23:49:19 Searching Annoy index using 1 thread, search_k = 3000
23:49:23 Annoy recall = 100%
23:49:24 Commencing smooth kNN distance calibration using 1 thread with target n_neighbors = 30
23:49:27 Initializing from normalized Laplacian + noise (using RSpectra)
23:49:27 Commencing optimization for 200 epochs, with 614028 positive edges
Using method 'umap'
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
23:49:34 Optimization finishedDimPlot(ipfatlas.xen1panel.epi.cells, group.by='Manuscript_Identity')
DimPlot(ipfatlas.xen1panel.epi.cells, group.by='Subclass_Cell_Identity')
DimPlot(ipfatlas.xen1panel.epi.cells, group.by='Subject_Identity')
DimPlot(ipfatlas.xen1panel.epi.cells, group.by='Disease_Identity')
Check fibrotic DATP expression
FeaturePlot(ipfatlas.xen1panel.epi.cells, features=c('CLDN4','KRT7','LGALS3','GDF15'), order=T, max.cutoff = 'q99', min.cutoff='q1')
FeaturePlot(ipfatlas.xen1panel.epi.cells, features=c('KRT17','KRT5','TP63','PDPN'), order=T, max.cutoff = 'q99', min.cutoff='q1')
Subset fibroblasts
ipfatlas.xen1panel.fib.cells <- subset(ipfatlas.xen1panel.cells, subset=CellType_Category=='Stromal')ipfatlas.xen1panel.fib.cells <- NormalizeData(ipfatlas.xen1panel.fib.cells)Normalizing layer: counts
Performing log-normalization
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.xen1panel.fib.cells <- FindVariableFeatures(ipfatlas.xen1panel.fib.cells)Finding variable features for layer counts
Calculating gene variances
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.xen1panel.fib.cells <- ScaleData(ipfatlas.xen1panel.fib.cells, features=rownames(ipfatlas.xen1panel.fib.cells))Centering and scaling data matrix
|
| | 0%
|
|=====================================================================================================================| 100%ipfatlas.xen1panel.fib.cells <- RunPCA(ipfatlas.xen1panel.fib.cells)Warning: The following 9 features requested have zero variance; running reduction without them: TNF, ATP1B1, ENAH, IRF3, GZMK, IFNA1, IL17A, ARL14, ASCL3PC_ 1
Positive: MFAP5, SFRP2, PLA2G2A, MEDAG, FBN1, DCLK1, CD34, RARRES1, SEMA3C, PDPN
APOD, PCOLCE2, PI16, THBS2, KLF4, F3, LGALS3, IGF1, CXCL14, CTSL
VEGFA, NFKB1, THY1, MYC, PIM1, COL1A1, RAMP2, IL33, LGALS3BP, XBP1
Negative: MYH11, DGKG, DES, PLN, HIGD1B, RGS5, KCNK3, LMOD1, ITGA1, CSPG4
TRPC6, LAMC3, RERGL, PDGFRB, LYVE1, CNN1, APOLD1, FKBP5, LGR6, NID1
FCMR, STAT4, ADAMTS1, SLIT3, INMT, CRISPLD2, SCN7A, CD4, NTRK2, P2RX1
PC_ 2
Positive: POSTN, COL8A1, COL1A1, COL5A2, CTHRC1, PDE4D, SCN7A, LTBP2, RORA, FOXN3
FSCN1, THY1, VCAM1, CD4, ITGA1, CSTA, WNT2, SPRY1, GJA5, SAMD3
ADAM17, FGFR4, MIS18BP1, IGFBP3, HIF1A, DIRAS3, TGFB1, EPCAM, SLC1A3, DPP6
Negative: MFAP5, PI16, PLA2G2A, PCOLCE2, CD34, MYC, RAMP2, NTN4, DGKG, MYH11
F3, MEDAG, NID1, DES, CXCL14, GNG11, ADAMTS1, IL6, RERGL, CRISPLD2
KLF4, NTRK2, PLN, HIGD1B, STEAP4, APOLD1, APOD, FABP3, RGS5, VEGFA
PC_ 3
Positive: BAIAP2L1, DGKG, CNN1, PDGFRB, MYH11, APOLD1, HIGD1B, ATF4, RGS5, LYVE1
PLN, PRDX6, THY1, HIF1A, LMOD1, ITGA1, DES, CSPG4, KCNK3, EGLN3
CD4, TGFB1, RERGL, ADAMTS1, XBP1, TNFAIP3, GPI, DUSP1, LGALS3BP, LAMC3
Negative: SVEP1, APOD, MFAP5, PI16, SCN7A, PLA2G2A, RORA, FBN1, CXCL14, DCLK1
TC2N, SFRP2, CD34, SEMA3C, PCOLCE2, RAMP2, CD38, F3, FGFR4, ASCL1
CLCA1, TCL1A, TP73, IL13, SMIM24, CXCR3, RNASE3, CX3CR1, LILRA4, FOXI1
PC_ 4
Positive: INMT, SCN7A, TRPC6, LAMC3, NID1, HIGD1B, SVEP1, FGFR4, TSPAN8, RGS5
KCNK3, SPRY1, STEAP4, ITGA1, FKBP5, APOD, RORA, CRISPLD2, EGFR, TC2N
PDE4D, SLIT3, IFI6, LGALS3BP, SLC1A3, FAM184A, CD4, STAT4, WNT2, PRDX6
Negative: PLN, DES, CNN1, CTHRC1, THBS2, POSTN, MYH11, COL1A1, RERGL, LGR6
WT1, IGFBP3, FSCN1, FABP3, CDKN2A, PDPN, TGFB1, MEDAG, VEGFA, SLC2A1
LMOD1, CD24, HIF1A, NTRK2, RBP4, CXCL13, P2RX1, LYVE1, COL8A1, DIRAS3
PC_ 5
Positive: IL6, CRISPLD2, IRF1, DUSP1, MYC, ADAMTS1, NFKB1, PIM1, KLF4, HIF1A
FGFR4, SLC1A3, ERN1, SCN7A, INMT, KLF5, SLC2A1, TNFAIP3, SVEP1, DES
DNAJB9, EGLN3, IRF8, FAM184A, VEGFA, PRDX6, SOX9, DPP6, LTBP2, RERGL
Negative: THY1, RGS5, LAMC3, HIGD1B, KCNK3, MFAP5, PI16, PCOLCE2, COL1A1, NTN4
PDGFRB, FBN1, LGALS3BP, GNG11, RAMP2, CD34, COL5A2, CTHRC1, FKBP5, TRPC6
CD24, IFI6, DCLK1, CXCL14, POSTN, WNT5A, IGFBP3, FABP3, CSPG4, CD4 ipfatlas.xen1panel.fib.cells <- RunUMAP(ipfatlas.xen1panel.fib.cells, dims=1:30)23:49:56 UMAP embedding parameters a = 0.9922 b = 1.112
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
23:49:56 Read 6430 rows and found 30 numeric columns
23:49:56 Using Annoy for neighbor search, n_neighbors = 30
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
23:49:56 Building Annoy index with metric = cosine, n_trees = 50
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
23:49:56 Writing NN index file to temp file /tmp/RtmpvuYL8i/filef3dd7728f3e9
23:49:56 Searching Annoy index using 1 thread, search_k = 3000
23:49:58 Annoy recall = 100%
23:49:59 Commencing smooth kNN distance calibration using 1 thread with target n_neighbors = 30
23:50:01 Initializing from normalized Laplacian + noise (using RSpectra)
23:50:01 Commencing optimization for 500 epochs, with 279884 positive edges
Using method 'umap'
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
23:50:09 Optimization finishedDimPlot(ipfatlas.xen1panel.fib.cells, group.by='Manuscript_Identity')
DimPlot(ipfatlas.xen1panel.fib.cells, group.by='Subclass_Cell_Identity')
DimPlot(ipfatlas.xen1panel.fib.cells, group.by='Subject_Identity')
DimPlot(ipfatlas.xen1panel.fib.cells, group.by='Disease_Identity')
Check fibroblast subtype markers
FeaturePlot(ipfatlas.xen1panel.fib.cells, features=c('COL1A1','CTHRC1','INMT','SCN7A'), order=T, max.cutoff = 'q99', min.cutoff='q1')
FeaturePlot(ipfatlas.xen1panel.fib.cells, features=c('PLA2G2A','PI16','MFAP5','WNT5A'), order=T, max.cutoff = 'q99', min.cutoff='q1')
Simulate Cosmx 1k panel
The “Universal” gene panel
cosmx.1Kpanel.genes <- read.csv('CosMx Human Universal Panel Gene Target List_sept23.csv', header = TRUE)$Geneipfatlas.cosmx1k.cells <- subset(ipfatlas.cells, features=cosmx.1Kpanel.genes)Some genes in the Cosmx panel are not in the IPF dataset. Many of these genes are listed with a “/” suggesting that the probe cannot distinguish these isoforms. For simplicity I am not going to clean up the gene list. So actual list is a bit shorter. These are the missing genes:
setdiff(cosmx.1Kpanel.genes, rownames(ipfatlas.cells)) [1] "BBLN" "C11orf96" "CALM2" "CCL15" "CCL3/L1/L3" "CCL4/L1/L2" "CD68" "CLEC4A"
[9] "CXCL1/2/3" "EIF5A/L1" "FCGR3A/B" "FKBP11" "FLT3LG" "H2AZ1" "H4C3" "HBA1/2"
[17] "HCAR2/3" "HILPDA" "HLA-DQB1/2" "HLA-DRB" "HSPA1A/B" "IFITM1" "IFNA1/13" "IFNL2/3"
[25] "IGKC" "INS" "KLRK1" "KRT6A/B/C" "LY75" "MAP1LC3B/2" "MHC I" "MIF"
[33] "MZT2A/B" "Negative01" "Negative02" "Negative03" "Negative04" "Negative05" "Negative06" "Negative07"
[41] "Negative08" "Negative09" "Negative10" "P2RX5" "PDGFRA" "PF4/V1" "PTPRCAP" "SAA1/2"
[49] "SERPINA3" "SOX2" "TAP2" "TNXA/B" "TPSAB1/B2" "XCL1/2" Standard clustering
ipfatlas.cosmx1k.cells <- NormalizeData(ipfatlas.cosmx1k.cells)Normalizing layer: counts
Performing log-normalization
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.cosmx1k.cells <- FindVariableFeatures(ipfatlas.cosmx1k.cells)Finding variable features for layer counts
Calculating gene variances
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.cosmx1k.cells <- ScaleData(ipfatlas.cosmx1k.cells, features=rownames(ipfatlas.cosmx1k.cells))Centering and scaling data matrix
|
| | 0%
|
|=====================================================================================================================| 100%ipfatlas.cosmx1k.cells <- RunPCA(ipfatlas.cosmx1k.cells)PC_ 1
Positive: TYROBP, FCER1G, PSAP, HLA-DRA, GPX1, MARCO, GLUL, ACP5, CSTB, APOC1
C1QA, C1QB, OLR1, VIM, MRC1, C1QC, AIF1, CD74, SERPINA1, LGALS3
HLA-DPA1, GPNMB, MSR1, CXCL16, HLA-DPB1, LGALS1, APOE, S100A6, MS4A4A, RBM47
Negative: MALAT1, FYN, CD69, ITK, CCL5, ETS1, IL32, STAT4, RUNX3, GZMB
PRF1, CLEC2D, NKG7, GZMA, GNLY, CST7, CD3D, KLRB1, SPOCK2, CD2
CD3E, RORA, IKZF3, YES1, GZMH, TOX, CTSW, CD3G, ITM2A, KLRF1
PC_ 2
Positive: SRGN, PTPRC, CXCR4, CD69, CCL5, STAT4, CD53, RUNX3, ITK, NKG7
ADGRE5, FYN, ARHGDIB, CST7, GZMB, CD3D, PRF1, DUSP2, GZMA, B2M
CD52, CD2, HCST, RGS1, GNLY, FYB1, TNFRSF1B, CD3E, KLRB1, GZMH
Negative: MGP, IGFBP7, SPARCL1, CALD1, CCDC80, COL1A2, RARRES2, EGFR, CAV1, CLU
LUM, PTK2, OSMR, COL5A2, COL1A1, COL6A1, DCN, COL3A1, IGFBP5, MYL9
COL6A2, BGN, CD59, SLPI, TM4SF1, COL6A3, COL4A1, KRT19, PLAC9, KRT18
PC_ 3
Positive: CDH1, KRT19, CLDN4, TACSTD2, AGR2, EPCAM, KRT8, CXCL17, PIGR, KRT18
SLPI, CD24, CELSR1, KRT7, ADGRV1, IL20RA, EPHA2, LCN2, DDR1, ERBB3
FGFR2, ITGA3, ALCAM, TNFRSF19, WIF1, GDF15, EFNA5, IL7, MET, ITGB6
Negative: COL1A2, LUM, COL3A1, COL6A2, DCN, COL6A3, BGN, COL6A1, MGP, COL1A1
SPARCL1, CALD1, COL5A2, RARRES2, COL5A1, MXRA8, IGFBP6, CDH11, PTGDS, MEG3
THBS2, TPM2, CACNA1C, DPT, PDGFRB, CXCL12, PTGIS, FGF7, MMP2, COL4A1
PC_ 4
Positive: B2M, CCL5, CD52, CTSW, IL32, CD3D, MYL12A, GZMB, GZMA, RPL32
FAU, GZMH, CD69, CRIP1, ARHGDIB, FABP4, CD3E, PRF1, RPL37, ITK
NKG7, SOD1, CD2, RPL34, CALM1, GNLY, RPL21, CST7, IL6ST, KLRB1
Negative: NLRP3, IL1R2, VCAN, HIF1A, NEAT1, THBS1, SAT1, ITGAX, CSF2RA, VEGFA
BASP1, TIMP1, ADGRE2, GPR183, SOD2, G0S2, MAP2K1, CD93, TLR2, CSF3R
MALAT1, RBM47, AREG, ST6GALNAC3, RGS2, CXCL8, CD14, IL1R1, S100A8, PPARD
PC_ 5
Positive: RPL21, RPL34, RPL32, RPL37, IL6ST, FAU, UBA52, RACK1, RPL22, GPR183
NACA, BTF3, AREG, ZFP36, ATP5F1E, RGS2, JUNB, TPT1, SRGN, TIMP1
RGS1, NLRP3, PFN1, CXCR4, B2M, SEC61G, IL1R2, HMGB2, VSIR, TPI1
Negative: FABP4, MSR1, INHBA, PPARG, C1QB, C1QA, APOC1, DST, C1QC, LGALS3BP
APOE, NUPR1, MARCO, OASL, FN1, IFI6, CD5L, ITGB8, CCL18, ACP5
IFI27, SLCO2B1, CXCL5, ACACB, FFAR4, MRC1, PDGFC, DDX58, GPNMB, DCN ipfatlas.cosmx1k.cells <- RunUMAP(ipfatlas.cosmx1k.cells, dims=1:30)00:18:10 UMAP embedding parameters a = 0.9922 b = 1.112
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
00:18:10 Read 253424 rows and found 30 numeric columns
00:18:10 Using Annoy for neighbor search, n_neighbors = 30
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
00:18:10 Building Annoy index with metric = cosine, n_trees = 50
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
00:18:51 Writing NN index file to temp file /tmp/RtmpvuYL8i/filef3dd3e1559c9
00:18:51 Searching Annoy index using 1 thread, search_k = 3000
00:20:42 Annoy recall = 100%
00:20:43 Commencing smooth kNN distance calibration using 1 thread with target n_neighbors = 30
00:20:57 Initializing from normalized Laplacian + noise (using RSpectra)
00:22:01 Commencing optimization for 200 epochs, with 11742532 positive edges
Using method 'umap'
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
00:24:24 Optimization finishedDimPlot(ipfatlas.cosmx1k.cells, group.by='CellType_Category')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.cosmx1k.cells, group.by='Manuscript_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.cosmx1k.cells, group.by='Subclass_Cell_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.cosmx1k.cells, group.by='Subject_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
DimPlot(ipfatlas.cosmx1k.cells, group.by='Disease_Identity')Rasterizing points since number of points exceeds 100,000.
To disable this behavior set `raster=FALSE`
Subset epithelium
ipfatlas.cosmx1k.epi.cells <- subset(ipfatlas.cosmx1k.cells, subset=CellType_Category=='Epithelial')ipfatlas.cosmx1k.epi.cells <- NormalizeData(ipfatlas.cosmx1k.epi.cells)Normalizing layer: counts
Performing log-normalization
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.cosmx1k.epi.cells <- FindVariableFeatures(ipfatlas.cosmx1k.epi.cells)Finding variable features for layer counts
Calculating gene variances
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.cosmx1k.epi.cells <- ScaleData(ipfatlas.cosmx1k.epi.cells, features=rownames(ipfatlas.cosmx1k.epi.cells))Centering and scaling data matrix
|
| | 0%
|
|=====================================================================================================================| 100%ipfatlas.cosmx1k.epi.cells <- RunPCA(ipfatlas.cosmx1k.epi.cells)Warning: The following 3 features requested have zero variance; running reduction without them: PDS5A, CTSD, CRPPC_ 1
Positive: MALAT1, CD44, CSF3R, BMP1, ROR1, SORBS1, LDLR, NFKBIA, ITGB6, IL1R1
LAMP3, COL8A1, FGFR1, TTN, DPP4, ETV5, OSMR, TNFRSF10B, EFNB2, BMP2
NDRG1, HIF1A, CD83, COL12A1, NFKB1, ITGA9, CXCL8, ADGRF5, VEGFA, SOD2
Negative: GSTP1, CALM1, CRIP1, B2M, CD24, SOD1, CD59, HSP90AA1, ATP5F1E, TUBB4B
IGFBP7, CSTB, PLAC8, RPL34, S100A6, FAU, IL6ST, LGALS3, HSP90AB1, RPL21
RPL32, UBA52, CLU, GSN, BASP1, IGFBP5, GLUL, TPT1, KRT8, PTGES3
PC_ 2
Positive: IL7, PLAC8, COL21A1, IGF1R, BASP1, PSD3, IL16, NELL2, IGFBP7, TNFRSF19
MAML2, CELSR1, RARB, CD24, IGFBP5, ALCAM, BRCA1, ATR, NRXN3, DLL1
CRIP1, WNT5B, ITGB8, PSCA, NPR2, HSP90AA1, CASP8, RGS5, ATM, TSHZ2
Negative: NFKBIA, TM4SF1, FABP5, IFITM3, RPL37, RACK1, TPT1, CXCL8, RPL32, CD44
LAMP3, ENO1, RPL22, CXCL17, RPL21, PHLDA2, NACA, SERPINA1, CD63, FAU
RPL34, DUSP6, UBA52, LDHA, ZFP36, MT1X, BTF3, AREG, MYC, BMP2
PC_ 3
Positive: LAMP3, CSF3R, SLPI, WIF1, TOX, FABP5, PIGR, CD36, SERPINA1, CXCL17
CD74, TTN, HLA-DPA1, FASN, HLA-DRA, CCL20, HLA-DPB1, BMP1, DMBT1, ETV5
CD83, CD38, CSF3, RBM47, SOD2, SPRY4, ADGRV1, AGR2, IL20RA, BMP2
Negative: DST, KRT17, CAV1, S100A2, SPOCK2, KRT15, MIR4435-2HG, KRT5, CALD1, KRT7
MYL9, IL32, COL4A2, CEACAM6, COL6A2, KRT19, CYTOR, COL4A1, PXDN, TIMP1
ANKRD1, ROR1, IGFBP6, NRG1, GAS6, TNNC1, AXL, COL17A1, TNFRSF12A, S100A10
PC_ 4
Positive: SPOCK2, TNNC1, MYL9, CAV1, ANKRD1, COL4A2, VEGFA, COL8A1, COL4A1, CEACAM6
IL32, S100A6, ROR1, B2M, IFI27, COL12A1, GAS6, S100A4, GSTP1, ATP5F1E
NUPR1, RGCC, MYL12A, S100A10, KRT7, RPL21, SLC40A1, CRIP1, CD63, IFI6
Negative: EFNA5, MET, KRT17, NEAT1, PSD3, SQSTM1, PTK2, GLUD1, S100A2, ITGB8
TNFRSF10B, KRT5, RORA, TNFRSF21, EPHA2, LYN, KRT15, MAML2, SMAD3, ARID5B
IL1R1, RBM47, SERPINB5, CDH1, STAT3, NFKB1, CTNNB1, SAT1, IL4R, SLC2A1
PC_ 5
Positive: CDH1, NEAT1, ALCAM, CD55, CFLAR, SPOCK2, TNNC1, COL4A2, COL12A1, ANKRD1
BCL2L1, ITGA3, LDLR, ITGB6, MECOM, MALAT1, IL32, CLDN4, LYN, COL4A1
STAT3, VEGFA, ATF3, BRAF, ADGRF5, CAV1, MYL9, KRT7, EFNA1, GADD45B
Negative: TIMP1, PTGIS, VCAM1, COL6A2, SLPI, RPL34, MGP, RPL32, COL1A2, RARRES1
CCL2, CLU, RPL21, SCGB3A1, TPM2, MT1X, NACA, IGFBP6, PTGDS, LTF
BTF3, THBS2, TPT1, LAMA4, FAU, IL6ST, COL3A1, MT2A, RPS4Y1, IL6 ipfatlas.cosmx1k.epi.cells <- RunUMAP(ipfatlas.cosmx1k.epi.cells, dims=1:30)00:24:57 UMAP embedding parameters a = 0.9922 b = 1.112
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
00:24:57 Read 13370 rows and found 30 numeric columns
00:24:57 Using Annoy for neighbor search, n_neighbors = 30
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
00:24:57 Building Annoy index with metric = cosine, n_trees = 50
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
00:24:59 Writing NN index file to temp file /tmp/RtmpvuYL8i/filef3dd24bad2d6
00:24:59 Searching Annoy index using 1 thread, search_k = 3000
00:25:02 Annoy recall = 100%
00:25:04 Commencing smooth kNN distance calibration using 1 thread with target n_neighbors = 30
00:25:06 Initializing from normalized Laplacian + noise (using RSpectra)
00:25:06 Commencing optimization for 200 epochs, with 607432 positive edges
Using method 'umap'
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
00:25:13 Optimization finishedDimPlot(ipfatlas.cosmx1k.epi.cells, group.by='Manuscript_Identity')
DimPlot(ipfatlas.cosmx1k.epi.cells, group.by='Subclass_Cell_Identity')
DimPlot(ipfatlas.cosmx1k.epi.cells, group.by='Subject_Identity')
DimPlot(ipfatlas.cosmx1k.epi.cells, group.by='Disease_Identity')
Check fibrotic DATP expression
Subset fibroblasts
ipfatlas.cosmx1k.fib.cells <- subset(ipfatlas.cosmx1k.cells, subset=CellType_Category=='Stromal')ipfatlas.cosmx1k.fib.cells <- NormalizeData(ipfatlas.cosmx1k.fib.cells)Normalizing layer: counts
Performing log-normalization
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.cosmx1k.fib.cells <- FindVariableFeatures(ipfatlas.cosmx1k.fib.cells)Finding variable features for layer counts
Calculating gene variances
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
Calculating feature variances of standardized and clipped values
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|ipfatlas.cosmx1k.fib.cells <- ScaleData(ipfatlas.cosmx1k.fib.cells, features=rownames(ipfatlas.cosmx1k.fib.cells))Centering and scaling data matrix
|
| | 0%
|
|=====================================================================================================================| 100%ipfatlas.cosmx1k.fib.cells <- RunPCA(ipfatlas.cosmx1k.fib.cells)Warning: The following 6 features requested have zero variance; running reduction without them: IL22RA1, LAIR1, ATG10, RBPJ, RAC1, MYL7PC_ 1
Positive: S100A10, DCN, CCDC80, MMP2, CFD, LUM, IGFBP6, ANXA2, S100A6, COL1A2
TIMP1, ANXA1, LGALS3, MFAP5, PTGIS, EMP3, RARRES1, VCAN, GPNMB, LDHA
CD63, VIM, COL6A1, GSN, PSAP, TSHZ2, CYP1B1, COL1A1, TPT1, RACK1
Negative: MALAT1, MYH11, COX4I2, NOTCH3, RGS5, ADGRF5, NDUFA4L2, ACTG2, COL4A1, CARMN
ESAM, CACNA1C, ITGA1, PTEN, RYR2, CSPG4, COL4A2, FKBP5, ACTA2, TNNT2
TPM2, CALD1, PDGFRB, ANGPT2, ZBTB16, PDGFA, ARHGDIB, TM4SF1, LYVE1, MYL9
PC_ 2
Positive: CFD, MFAP5, CCDC80, SLPI, DCN, APOD, PTGIS, IGFBP6, CLU, CD34
IGF2, ACKR3, TSHZ2, ITM2A, CD55, SOD2, FGFR1, NPR1, CXCL14, ABL2
MEG3, GSN, VEGFA, ESR1, TGFBR2, DDR2, GPNMB, RSPO3, EGFR, TWIST2
Negative: MYL9, TPM2, BGN, NDUFA4L2, COX4I2, ACTA2, NOTCH3, CALD1, CAV1, RGS5
LGALS1, IGFBP7, TSC22D1, PFN1, ARHGDIB, ACTG2, ITGA1, CALM1, COL4A1, MYH11
ESAM, ADGRF5, TPM1, CACNA1C, PDGFRB, HSPB1, COL4A2, TPI1, NACA, VIM
PC_ 3
Positive: COL6A3, CDH11, COL3A1, COL1A1, COL5A2, COL8A1, ITGA8, FN1, BMP5, COL1A2
RYR2, COL5A1, LUM, ROR1, RORA, FAP, MMP2, PTGDS, PSD3, DST
TNFRSF19, TSHZ2, COL16A1, EPHA3, ENTPD1, VCAN, IL16, LDB2, ITGB8, ANGPT1
Negative: GPX3, IGFBP6, SLPI, MFAP5, MT2A, GSN, IGFBP5, MYC, ZFP36, LDHA
ACKR3, CDKN1A, RPL21, CRIP1, MYH11, JUNB, CLU, CRYAB, CD34, NDUFA4L2
IGFBP7, COX4I2, MT1X, ITM2A, FAU, IGF2, RAMP2, GADD45B, ACTG2, IFI27
PC_ 4
Positive: MGP, NUPR1, DCN, ITM2B, SPARCL1, CFD, S100A6, LUM, RPL34, RPL21
CLU, PLAC9, CD63, RARRES2, B2M, RPL32, IL16, GSN, LGALS1, S100A4
ITM2A, APOD, IFITM3, CXCL14, GPX3, GSTP1, UBA52, FAU, PTGDS, SLC40A1
Negative: MIR4435-2HG, CYTOR, THBS1, CD44, ABL2, NFKB1, HIF1A, VEGFA, TNFRSF10B, CDKN1A
IL1R1, SOD2, ITGAV, NEAT1, ETS1, LDLR, TNFRSF12A, MAP2K1, FGFR1, ATF3
GADD45B, MYC, DUSP5, IL6, LIF, CCL2, BAG3, PPARD, FGF2, STAT3
PC_ 5
Positive: PTGDS, CCL2, RARRES2, ZFP36, SAT1, JUNB, SOD2, MT2A, MT1X, GADD45B
IL6ST, MGP, RGCC, RPL34, FOS, RPL21, CXCL12, NFKBIA, IL6, RPL32
TNFSF13B, DUSP1, ANGPT1, SELENOP, THBS1, LUM, SRGN, ITGA2, ITGA8, JUN
Negative: MFAP5, COL12A1, COL1A2, IGFBP5, COL3A1, PTGIS, IGFBP6, ACKR3, SLPI, CRIP1
COL1A1, CD34, THBS2, COL5A1, TGFBI, CD55, IFI27, APP, MEG3, PTK2
IGF2, CALD1, CD276, BMP1, COL6A1, MMP14, SH3BGRL3, PDGFRB, COL6A2, GSN ipfatlas.cosmx1k.fib.cells <- RunUMAP(ipfatlas.cosmx1k.fib.cells, dims=1:30)00:25:30 UMAP embedding parameters a = 0.9922 b = 1.112
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
00:25:30 Read 6430 rows and found 30 numeric columns
00:25:30 Using Annoy for neighbor search, n_neighbors = 30
Found more than one class "dist" in cache; using the first, from namespace 'spam'
Also defined by ‘BiocGenerics’
00:25:30 Building Annoy index with metric = cosine, n_trees = 50
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
00:25:31 Writing NN index file to temp file /tmp/RtmpvuYL8i/filef3dd5dd09376
00:25:31 Searching Annoy index using 1 thread, search_k = 3000
00:25:33 Annoy recall = 100%
00:25:34 Commencing smooth kNN distance calibration using 1 thread with target n_neighbors = 30
00:25:36 Initializing from normalized Laplacian + noise (using RSpectra)
00:25:36 Commencing optimization for 500 epochs, with 275020 positive edges
Using method 'umap'
0% 10 20 30 40 50 60 70 80 90 100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
00:25:44 Optimization finishedDimPlot(ipfatlas.cosmx1k.fib.cells, group.by='Manuscript_Identity')
DimPlot(ipfatlas.cosmx1k.fib.cells, group.by='Subclass_Cell_Identity')
DimPlot(ipfatlas.cosmx1k.fib.cells, group.by='Subject_Identity')
DimPlot(ipfatlas.cosmx1k.fib.cells, group.by='Disease_Identity')
Check fibroblast subtype markers
FeaturePlot(ipfatlas.cosmx1k.fib.cells, features=c('COL1A1','CTHRC1','INMT','SCN7A'), order=T, max.cutoff = 'q99', min.cutoff='q1')Warning: The following requested variables were not found: CTHRC1, INMT, SCN7A
FeaturePlot(ipfatlas.cosmx1k.fib.cells, features=c('PLA2G2A','PI16','MFAP5','WNT5A'), order=T, max.cutoff = 'q99', min.cutoff='q1')Warning: The following requested variables were not found: PLA2G2A, PI16
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